Metal-oxide Memristive Devices Research Articles

Neuronal Multi Unit Activity Processing with Metal Oxide Memristive Devices

AbstractIntra‐cortical brain‐machine interfaces (BMIs), able to decode neural activity in real‐time, represent a revolutionary opportunity for treating medical conditions. However, traditional systems focusing on single‐neuron spike detection require high processing rates and power, hindering the up‐scaling for neurons‐population monitoring in clinical application. An intriguing proposition is the memristive integrating sensor (MIS) approach, which uses resistive RAM (RRAM) for threshold‐based neural activity detection. MIS leverages analogue multi‐state switching properties of metal‐oxide RRAM to compress neural inputs by encoding above‐threshold events in resistance displacement, facilitating efficient data down‐sampling in the post‐processing, enabling low‐power, high‐channel systems. Initially tested on spikes and local field potentials, here MIS is adapted to process multi‐unit activity envelope (eMUA)—the envelope of entire spiking activity—which has recently been proposed as crucial input for real‐time neuro‐prosthetic control. Prior necessary modifications to the MIS for effective operation, this adaptation achieved over 95% sensitivity across two types of metal‐oxide devices: Pt/TiOx/Pt and TiN/HfOx/TiN, proving its platform‐agnostic capabilities. Furthermore, towards the integration of MIS with silicon chips, it is shown that it can reduce total system power consumption to below 1 µW, as RRAM encoding stage relaxes the signal preservation and noise requirements that challenge traditional complementary metal‐oxide‐semiconductor (CMOS) front‐ends. This eMUA‐MIS adaptation offers a viable pathway for developing more scalable and efficient BMIs for clinical use.

Effect of Proton Irradiation Fluence on the Linearity and Symmetry of Conductance Tuning of a HfO x /TiO x Heterojunction-Based Memristor

Binary metal oxide memristor-based artificial neural synapses offer the advantages of high-density information storage, high-efficiency computing, low power consumption and strong antiradiation damage and are therefore promising aerospace electronic devices. However, the unclear damage trends and mechanisms causing nonlinearity and asymmetry during conductance tuning under irradiation have restricted further development. Here, the damage trends and mechanisms caused by proton irradiation in Au/HfO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>x</i></sub> /TiO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>x</i></sub> /Ti memristors were studied. Results show that when the fluence of proton (25 MeV) irradiation increases to 2×10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">11</sup> p/cm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> , the Au/HfO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>x</i></sub> /TiO <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"><i>x</i></sub> /Ti memristor cannot return to its original high-resistance state upon conductance reduction, making the conductance tuning less linear and symmetric. X-ray photoelectron spectroscopy analysis revealed that the mechanism underlying the decrease in the linearity and symmetry of the conductance tuning is proton irradiation increasing oxygen vacancies in the heterojunction. These results can guide better design of metal oxide memristive devices suitable for aerospace devices and better modeling of devices with nonideal properties under irradiation.

AI Acceleration Enabled by Nanoelectronic Memristive Devices

Here we present an analysis of the current state in the field of development of hardware accelerators of artificial intelligence (AI). Despite the fairly good progress made over the past decades, this area is experiencing a number of significant difficulties in its development. The solution to this problem lies in the application of new approaches to the organization of computing, in particular, computing in memory enabled by nanoelectronic memristive devices. We provide an overview of state-of-art systems, as well as our own version of the experimental concept of AI accelerators based on metal-oxide memristive devices and the massively parallel architecture for information processing.

Neurohybrid Memristive CMOS-Integrated Systems for Biosensors and Neuroprosthetics.

Here we provide a perspective concept of neurohybrid memristive chip based on the combination of living neural networks cultivated in microfluidic/microelectrode system, metal-oxide memristive devices or arrays integrated with mixed-signal CMOS layer to control the analog memristive circuits, process the decoded information, and arrange a feedback stimulation of biological culture as parts of a bidirectional neurointerface. Our main focus is on the state-of-the-art approaches for cultivation and spatial ordering of the network of dissociated hippocampal neuron cells, fabrication of a large-scale cross-bar array of memristive devices tailored using device engineering, resistive state programming, or non-linear dynamics, as well as hardware implementation of spiking neural networks (SNNs) based on the arrays of memristive devices and integrated CMOS electronics. The concept represents an example of a brain-on-chip system belonging to a more general class of memristive neurohybrid systems for a new-generation robotics, artificial intelligence, and personalized medicine, discussed in the framework of the proposed roadmap for the next decade period.

Integrated analog neurons inspired by mimicking synapses with metal-oxide memristive devices

Artificial neural networks (NNs) integrated with emerging devices are one promising approach to overcoming the limitations of artificial intelligence hardware based on existing CMOS technologies. Two-terminal memristive devices consisting of metal-oxide WOx/MgO were fabricated and investigated in order to add synaptic memory functions to NN hardware. The device showed analog conductance changes similar to a biological synapse’s spike-timing dependent plasticity as well as its binarized characteristics, which depend on waveforms of voltage spikes. To exploit such properties of the memristive devices for practical hardware, NN simulations were executed taking into account analog and binary synapses, resulting in good accuracy. For rapid NN prototyping with resistive synapses, we developed an analog-to-digital mixed circuit with variable resistors, which was fabricated using a standard CMOS process. Thirty-two analog neurons, where 1 neuron comprises 32 synapses, performed 1k synaptic operations at approximately 10 mW and generate neuronal firing at approximately less than 2 μs.

Flexible Monte-Carlo approach to simulate electroforming and resistive switching in filamentary metal-oxide memristive devices

An original approach has been presented to model the regularities and parameters of resistive switching based on the kinetic Monte Carlo (kMС) 3D simulation of stochastic migration of oxygen vacancies/ions in metal-oxide memristive devices promising for applications in emerging nonvolatile memory, in-memory and neuromorphic computing systems. The efficiency and flexibility of the approach is demonstrated by the examples of experimentally realized Au/oxide/TiN memristive device structures, in which yttria-stabilized zirconia (ZrO2(Y)) polycrystalline films and amorphous SiOx columnar films obtained by magnetron sputtering are used as the switching oxide material. The proposed approach combines the universal kMС framework for ion migration and relatively simple physics-based methods for taking into account additional factors related to the structure and morphology of specific oxide material, interface phenomena and energetics of electronic processes, and therefore does not require time-consuming calculations and is not critical to the computing power.

Examining Oxygen Exchange Kinetics of Anomalous Resistive Switching By Optical Tracking in Redox Based Memristive Devices

Anomalous resistive switching in redox based memristive devices has recently gained increased attention due to compelling demonstrations of its origin related to homogenous oxygen exchange with device electrodes [1]. Insights into this mechanism are expected to improve understanding of the origins of the competing switching mechanism in mixed valent metal oxide memristive devices. Previously, open systems were shown to be susceptible to oxygen exchange at triple phase boundaries[2] and to the incorporation of humidity,[3,4] under applied voltages, from the gas environment. These environmental exchanges can result in modified space charge layers and surface defect chemistries that in turn, can lead to modified resistive switching properties. Strategies to mitigate these effects have relied on the addition of an interlayer acting as a source of oxygen defects, also known as an exchange layer [5–7]. While exchange between the switching layer and the interlayer has been identified as the origin of resistance change, fundamental aspects of the exchange kinetics remain unclear. Vacuum based techniques, relying on compositional analysis such as EDS in TEM, have been used to demonstrate changes in chemistry, but provided limited information about the details of the exchange kinetics. Being able to track the dynamics of oxygen migration through the metal oxide semiconductor volume, and exchange at the interface over the entire area of the device, could provide insights into the kinetics and thermodynamic limitations responsible for switching speed and device retention. Mastering these two properties is fundamental for the development of anomalous resistive switching elements with increased stability, faster switching speeds and larger hysteresis. To gain further insight into the oxygen exchange kinetics, we investigated the resistive switching properties of a bilayer oxide system based on PrxCe1-xO2/CeO2 stacks, sandwiched between transparent electrodes. Our study focuses on the model material, PrxCe1-xO2 (PCO), whose defect chemical and transport properties have been previously studied both in bulk and thin film forms[8,9]. We leverage the existence of an interband gap optical absorption, related to the oxidation state of Pr ions, to monitor the changes in the film’s non-stoichiometry. By carefully processing the sandwiched PCO, it is possible to quench in different degrees of oxygen non-stoichiometry to room temperature, as confirmed by optical absorption and Raman spectroscopy. By engineering the defect chemistry of PCO and using CeO2 as an exchange layer, we investigate and correlate the switching dynamics via electrical and optical measurements to gain insight into the speed and the amount of oxygen exchanged between the two layers. [1] D. Cooper, C. Baeumer, N. Bernier, A. Marchewka, C. La Torre, R. E. Dunin-Borkowski, S. Menzel, R. Waser, R. Dittmann, Adv. Mater. 2017, 29, 1. [2] C. Lenser, M. Patt, S. Menzel, A. Köhl, C. Wiemann, C. M. Schneider, R. Waser, R. Dittmann, Adv. Funct. Mater. 2014, 24, 4466. [3] M. Lübben, S. Wiefels, R. Waser, I. Valov, Adv. Electron. Mater. 2017, 1700458, 1700458. [4] E. Sediva, W. J. Bowman, J. C. Gonzalez-Rosillo, J. L. M. Rupp, Adv. Electron. Mater. 2018, 1800566, 1. [5] M. Ismail, Shafqat-Un Nisa, T. Akbar, A. M. Rana, Jinju Lee, S. Kim, 2019, 012101. [6] J. C. Gonzalez-rosillo, R. Ortega-hernandez, B. Arndt, M. Coll, 2019, 1800629, 1. [7] Z. Alamgir, J. Holt, K. Beckmann, N. C. Cady, Semicond. Sci. Technol. 2018, 33. [8] S. R. Bishop, T. S. Stefanik, H. L. Tuller, Phys. Chem. Chem. Phys. 2011, 13, 10165. [9] J. J. Kim, S. R. Bishop, N. Thompson, Y. Kuru, H. L. Tuller, Solid State Ionics 2012, 225, 198.

Oxygen Exchange Processes between Oxide Memristive Devices and Water Molecules.

Resistive switching based on transition metal oxide memristive devices is suspected to be caused by the electric-field-driven motion and internal redistribution of oxygen vacancies. Deriving the detailed mechanistic picture of the switching process is complicated, however, by the frequently observed influence of the surrounding atmosphere. Specifically, the presence or absence of water vapor in the atmosphere has a strong impact on the switching properties, but the redox reactions between water and the active layer have yet to be clarified. To investigate the role of oxygen and water species during resistive switching in greater detail, isotope labeling experiments in a N2 /H218 O tracer gas atmosphere combined with time-of-flight secondary-ion mass spectrometry are used. It is explicitly demonstrated that during the RESET operation in resistive switching SrTiO3 -based memristive devices, oxygen is incorporated directly from water molecules or oxygen molecules into the active layer. In humid atmospheres, the reaction pathway via water molecules predominates. These findings clearly resolve the role of humidity as both oxidizing agent and source of protonic defects during the RESET operation.

Artificial neuron operations and spike-timing-dependent plasticity using memristive devices for brain-inspired computing

The use of memristive devices for creating artificial neurons is promising for brain-inspired computing from the viewpoints of computation architecture and learning protocol. We present an energy-efficient multiplier accumulator based on a memristive array architecture incorporating both analog and digital circuitries. The analog circuitry is used to full advantage for neural networks, as demonstrated by the spike-timing-dependent plasticity (STDP) in fabricated AlOx/TiOx-based metal–oxide memristive devices. STDP protocols for controlling periodic analog resistance with long-range stability were experimentally verified using a variety of voltage amplitudes and spike timings.

Multibit memory operation of metal-oxide bi-layer memristors

Emerging nanoionic memristive devices are considered as the memory technology of the future and have been winning a great deal of attention due to their ability to perform fast and at the expense of low-power and -space requirements. Their full potential is envisioned that can be fulfilled through their capacity to store multiple memory states per cell, which however has been constrained so far by issues affecting the long-term stability of independent states. Here, we introduce and evaluate a multitude of metal-oxide bi-layers and demonstrate the benefits from increased memory stability via multibit memory operation. We propose a programming methodology that allows for operating metal-oxide memristive devices as multibit memory elements with highly packed yet clearly discernible memory states. These states were found to correlate with the transport properties of the introduced barrier layers. We are demonstrating memory cells with up to 6.5 bits of information storage as well as excellent retention and power consumption performance. This paves the way for neuromorphic and non-volatile memory applications.

The Role of Oxygen Vacancy Mobility at the Oxide Bulk and Electrode Interfaces for Ceria-Based Memristive Devices

Resistive switching is a recently extensively studied phenomenon for use in non-volatile memories as promising building blocks of future electronics. In such devices transition metal oxides are operated at high electric field strengths to initiate non-linear and hysteretic current-voltage relations that can be addressed in their various resistance states by voltage pulse amplitude or polarity. Despite the known importance of oxygen vacancy defects in metal-oxide memristive devices, systematic strategies on tuning their concentration and mobility in the transition metal oxide are rare. In this work, we utilized the fluorite-type ceria-gadolinia binary system Ce1-xGdxO2-x/2 (GDC) to tune the oxygen vacancy concentration in order to correlate structure and defect chemistry to the memristive properties. The mixed conduction of GDC is well investigated at high temperature, and it is possible to extrinsically dope over the wide range of 3 to 30 mol% Gd; however, has only recently been reported for ambient and high electric fields1. Therefore, we can tune the oxygen vacancy concentration, migration and association energy by the acceptor dopant as a model system to directly investigate the influence on resistive switching. For this purpose Pt-GDC-Pt sandwich structures were fabricated as resistive switching devices with cross-bar electrodes. Based on hysteretic current-voltage profiles, a maximum in resistive switching (with ROFF/RON up to 100) is found for the best ion conducting2 20 mol% Gd-doped CeO2. In-operando Raman spectroscopy revealed a homogeneous reduction of the ceria film during electroforming, together with a strong increase in conductivity. Additional high energy XPS measurements we observe the electrostatic potential drop at the electrode-oxide interface. These experimental results directly prove that the electronic charge carriers generated by the formation and redistribution of oxygen vacancies at high electric fields is the origin of memristance in doped ceria films, and that the switching happens in a rather homogeneous manner. The importance of oxygen vacancy migration in our model is further verified by the doping study. Doping levels of 10-20 mol% resulted in good switching properties, whereas for low (<10 mol%) and high (>20 mol%) concentrations no resistive switching is observed. In both cases the number of mobile oxygen vacancies is not sufficient, either because of the low extrinsic doping level, or because of the formation of bixbyite-like domains with ordered or clustered "immobile" oxygen vacancies. This clustering is observed by the appearance of new vibrational modes in Raman spectroscopy due to the reduced symmetry in a defective crystal. Excitingly, one can study based on the findings the direct implication of "free" oxygen vacancies at low and high mobilites and also of "trapped" clustered vacancies on both, resistive switching and also local structural symmetry breaks3. In conclusion, we observed a clear correlation of oxygen vacancy mobility and switching characteristics. We suggest that a high concentration of mobile oxyen vacancies is of general importance when designing oxide materials for future computing applications.

Unsupervised learning in probabilistic neural networks with multi-state metal-oxide memristive synapses.

In an increasingly data-rich world the need for developing computing systems that cannot only process, but ideally also interpret big data is becoming continuously more pressing. Brain-inspired concepts have shown great promise towards addressing this need. Here we demonstrate unsupervised learning in a probabilistic neural network that utilizes metal-oxide memristive devices as multi-state synapses. Our approach can be exploited for processing unlabelled data and can adapt to time-varying clusters that underlie incoming data by supporting the capability of reversible unsupervised learning. The potential of this work is showcased through the demonstration of successful learning in the presence of corrupted input data and probabilistic neurons, thus paving the way towards robust big-data processors.

State of the art of metal oxide memristor devices

AbstractMemristors are one of the emerging technologies that can potentially replace state-of-the-art integrated electronic devices for advanced computing and digital and analog circuit applications including neuromorphic networks. Over the past few years, research and development mostly focused on revolutionizing the metal oxide materials, which are used as core components of the popular metal-insulator-metal memristors owing to their highly recognized resistive switching behavior. This paper outlines the recent advancements and characteristics of such memristive devices, with a special focus on (i) their established resistive switching mechanisms and (ii) the key challenges associated with their fabrication processes including the impeding criteria of material adaptation for the electrode, capping, and insulator component layers. Potential applications and an outlook into future development of metal oxide memristive devices are also outlined.